On 8/29/25 00:47, Peter Maydell wrote:
On Thu, 28 Aug 2025 at 13:10, Richard Henderson <richard.hender...@linaro.org> wrote:Make use of the fact that target_elf_gregset_t is a proper structure. Drop ELF_NREG, target_elf_greg_t, and tswapreg. Signed-off-by: Richard Henderson <richard.hender...@linaro.org> --- linux-user/arm/target_elf.h | 11 +++++++---- linux-user/arm/elfload.c | 8 +++----- 2 files changed, 10 insertions(+), 9 deletions(-) diff --git a/linux-user/arm/target_elf.h b/linux-user/arm/target_elf.h index 94db3738e8..fa8f8af2f3 100644 --- a/linux-user/arm/target_elf.h +++ b/linux-user/arm/target_elf.h @@ -8,16 +8,19 @@ #ifndef ARM_TARGET_ELF_H #define ARM_TARGET_ELF_H +#include "target_ptrace.h" + #define HAVE_ELF_HWCAP 1 #define HAVE_ELF_HWCAP2 1 #define HAVE_ELF_PLATFORM 1 #define HAVE_ELF_CORE_DUMP 1 -typedef abi_ulong target_elf_greg_t; - -#define ELF_NREG 18 +/* + * See linux kernel: arch/arm/include/asm/elf.h, where + * elf_gregset_t is mapped to struct pt_regs via sizeof. + */ typedef struct target_elf_gregset_t { - target_elf_greg_t regs[ELF_NREG]; + struct target_pt_regs pt; } target_elf_gregset_t; #endif diff --git a/linux-user/arm/elfload.c b/linux-user/arm/elfload.c index 47fe16a1a6..726d3ec25c 100644 --- a/linux-user/arm/elfload.c +++ b/linux-user/arm/elfload.c @@ -201,13 +201,11 @@ const char *get_elf_platform(CPUState *cs) #undef END } -#define tswapreg(ptr) tswapal(ptr) - void elf_core_copy_regs(target_elf_gregset_t *r, const CPUARMState *env) { for (int i = 0; i < 16; ++i) { - r->regs[i] = tswapreg(env->regs[i]); + r->pt.regs[i] = tswapal(env->regs[i]); } - r->regs[16] = tswapreg(cpsr_read((CPUARMState *)env)); - r->regs[17] = tswapreg(env->regs[0]); /* XXX */ + r->pt.cpsr = tswapal(cpsr_read((CPUARMState *)env)); + r->pt.orig_r0 = tswapal(env->regs[0]);Why is it OK to drop the "XXX" comment here ?
I assumed XXX meant "what is this", and the answer is orig_r0. I'm not even sure the value is wrong as-is, due to the way we process syscalls. r~
